We used four different measurements of agreement of eQTL effects when comparing different brain regions or tissues: AC, π1, Rb and caFC92. Each of these measures evaluates different aspects of replication: AC is an indication of the proportion of effects that have a shared direction of effect within the set of eQTLs that is significant in both the discovery and replication datasets and is expected to be 50% for random eQTL effects, π193 estimates the proportion of eQTL effects that are true positive in the replication cohort but does not take into account the effect direction and can be dependent on the replication dataset sample size, Rb94 effectively estimates the correlation between the eQTL-effect slopes (for example, β values from linear regression) while controlling for potential covariance in standard errors of those slopes and caFC measures the correlation between estimates of the fold change in expression values between alleles92. We note that whereas AC, π1 and Rb can be calculated from summary statistics, caFC requires access to genotype and expression data. We therefore limited the caFC analysis to comparisons within
